Ecological Interface Design for Knee and Hip Automatic Physiotherapy Assistant and Rehabilitation System Yeti Li Department of Systems Design Engineering, University of Waterloo Waterloo, Ontario, Canada Catherine M. Burns Department of Systems Design Engineering, University of Waterloo Waterloo, Ontario, Canada Dana Kulić Department of Electrical and Computer Engineering, University of Waterloo Waterloo, Ontario, Canada In this paper, we report on a recent interface design and evaluation process for a new knee and hip automatic physiotherapy assistant and rehabilitation system (ARS). Interface design was concurrent with the development of ARS. The ARS has the potential to improve the automation of rehabilitation treatments, by providing quantitative measures of a patient’s motion. However, the complexity of rehabilitation information available to the therapist has increased with this additional information. We applied Ecological Interface Design (EID) to understand the domain of physiotherapy and the role of the automation. Results of a Work Domain Analysis (WDA) revealed new functions and constraints in rehabilitation now accessible through the ARS, and provided the design requirements for interface design. A novel interface was designed which is currently undergoing evaluation to see if it improves the quality and experience of physiotherapy. This study provides an example of the advantages of using EID at the early phase of design, and how to apply EID to a system of increasing automaticity. INTRODUCTION In 2007, there were 332,000 hip and 719,000 knee replacement surgeries performed in the United States (National Hospital Discharge Survey, 2010) and these numbers are increasing. With more patients requiring rehabilitation, an accurate assessment of each patient is essential to manage patient workload. However, currently this assessment is performed by visual observation, patient questionnaires and goniometry (measurement of range of motion of a single joint while the patient is at rest). Even when carefully performed, there is an element of subjectivity and human error possible with these methods. An automatic physiotherapy assistant and rehabilitation system (ARS) is being developed by University of Waterloo and Cardon Rehabilitation & Medical Equipment Ltd. to provide automated assessment and monitoring using human motion tracking technology. This new technology should enhance pre- and post-rehabilitation assessment reliability by providing more detailed and accurate information. Specifically, an ARS provides a sensor measured view of the patient from which other quantitative measures can be derived. This should allow therapy to be monitored more accurately, improving rehabilitation therapy and shortening hospitalization. However, ARS has increased the amount and complexity of rehabilitation information available to the therapist. To understand this environment and to develop a user interface, a user interface design approach for complex systems, known as Ecological Interface Design (EID), has been applied to the design of the user interface for this system (Burns & Hajdukiewicz, 2004). More recently, EID has been used successfully in healthcare environments (Davies et al, 2005; Wu et al, 2012; Gorges et al, 2013). To our knowledge, automation in the rehabilitation setting has not been explored in the previous EID applications. The ARS interface design should be improved by using EID which, as a result, should help clarify the constraints and functions of the rehabilitation process. METHODS We began our study with a literature review on physiotherapy to understand basic principles related to knee and hip rehabilitation. An ethnographic observation of physiotherapists and patients was conducted at a local health center to understand the current process of therapy without the presence of ARS. This observation provided insight into how the ARS can best assist with the current therapy workflow. The manager of the rehabilitation services of the health center was interviewed to understand how out-patient information was controlled, which gave us insight into the presentation of each patient’s data. Our EID started from setting up the system boundary for this field. A Work Domain Analysis (WDA) was built, which described the components of this system independently from any tasks or events. Using a task independent analysis was important, because ARS is a new technology that has not yet been implemented in the current rehabilitation practice and therefore the tasks were unknown. A part-whole model was used to decompose the system into subsystems and